pacman::p_load(tidyverse,
here,
janitor,
ggthemes,
patchwork,
gghalves
)Datenvisualisierung 4
Multi-part plots and customisations
Lesungen
For further reading and practice on this topic, I suggest reading Section 11.5 (Communication: Themes) in Wickham et al. (2023), and Chapter 4 (Representing summary statistics) in Nordmann et al. (2022).
Learning objectives
In this section we will learn to
- build multi-part plots
- adjust the position of geoms
- customise our plots for better data communicate
Set-up
Packages
Today, we’re loading our relevant tidyverse packages directly: dplyr and ggplot. These are the only To aid us in loading in our data, we’re also loading the here package, and the janitor package which is useful for tidying up our data (e.g., the clean_names() function). To customise our plots we’re also using the ggthemes and patchwork packages. The former helps us produce plots that are colour-blind friendly, while the latter allows us to print multiple plots together. We also need a new package: gghalves.
Data
We’re again working with our slightly altered version of the english dataset from the languageR package.
df_eng <- read_csv(
here(
"daten",
"languageR_english.csv"
)
) |>
clean_names() |>
rename(
rt_lexdec = r_tlexdec,
rt_naming = r_tnaming
)1 Dodged density plots
We can produce density plots mapped along a catgorical variable by using geom_half_violin() from the gghalves package.
df_eng %>%
ggplot() +
aes(x = age_subject, y = rt_lexdec) +
geom_half_violin(alpha = .8)
gghalves::geom_half_violin()
1.1 Adding a boxplot
We can also add another geom to add more information to the plot. Let’s add a boxplot.
df_eng %>%
ggplot() +
aes(x = age_subject, y = rt_lexdec) +
geom_half_violin(alpha = .8) +
geom_boxplot()
1.2 position_nudge()
Maybe we want to move the boxplot so that it’s not overtop of the density plots, and so that it’s not quite as wide. We can do this by setting position to position_nudge(), and width to some value smaller than .75, which is the default width.
df_eng %>%
ggplot() +
aes(x = age_subject, y = rt_lexdec) +
geom_half_violin(alpha = .8) +
geom_boxplot(width = .3, # make less wide
position = position_nudge(x=0.2)
)
1.3 position_jitter() for scatterplots
This is from a family of options that allow us to alter the position of geoms. For example, Abbildung 4 A and B both show the exact same data, but Abbildung 4 B includes position = position_jitter(0.2) to move overlapping points. This way we get a good idea of how many observations there were across reaction times (y-axis).
Code
df_eng |>
ggplot() +
aes(x = age_subject, y = rt_lexdec) +
geom_point() +
labs(title = "geom_point()") +
df_eng |>
ggplot() +
aes(x = age_subject, y = rt_lexdec) +
geom_point(position = position_jitter(0.2),
alpha = 0.2)+
labs(title = "geom_point(position = position_jitter(0.2))") +
plot_annotation(tag_levels = "A")
1.4 Combining all three
If we put all of these plots together, we get a Abbildung 5.
Code
fig_no_colour <-
df_eng %>%
ggplot() +
aes(x = age_subject, y = rt_lexdec) +
geom_point(position = position_jitter(0.2),
alpha = 0.2) +
geom_half_violin() +
geom_boxplot(
outlier.shape = NA,
width = .3,
position = position_nudge(x=0.2)) 
2 Positioning errorbar plots
In the second report, you produced errorbar plots, but the errorbars were overlapping.
2.1 pivot_longer() |> summarise()
Let’s reproduce something similar using the english dataset. First, we’ll use pivot_longer() to lengthen our data, then we’ll create a summary of reaction times for the lexical decision task and naming task per age group.
sum_eng <-
df_eng |>
pivot_longer(
cols = c(rt_lexdec, rt_naming),
names_to = "task",
values_to = "rt"
) |>
summarise(
mean = mean(rt, na.rm = T),
sd = sd(rt, na.rm = T),
.by = c(age_subject, task)
) |>
mutate(age_subject = factor(age_subject, levels = c("young", "old"))) 2.2 Overlapping errorbars
If we create an errorbar plot of this data, we get Abbildung 7.
sum_eng |>
ggplot() +
aes(x = age_subject, y = mean, colour = task, shape = task) +
geom_point() +
geom_errorbar(aes(ymin = mean-sd, ymax = mean+sd))2.3 position_dodge()
We can add position = position_dodge(0.2) to force the errorbars to not overlap. We’ll also adjust their width so they’re not so wide (any value lower than 0.75).
sum_eng |>
ggplot() +
aes(x = age_subject, y = mean, colour = task, shape = task) +
geom_point() +
geom_errorbar(aes(ymin = mean-sd, ymax = mean+sd),
position = position_dodge(0.2),
width = 0.2)
2.4 dodging all relevant geoms
But now we’ve left the points behind. We need to also dodge the points, so we add position_dodge() to geom_point(), making sure to use the same value as we did with geom_errorbar().
sum_eng |>
ggplot() +
aes(x = age_subject, y = mean, colour = task, shape = task) +
geom_point(position = position_dodge(0.2)) +
geom_errorbar(aes(ymin = mean-sd, ymax = mean+sd),
position = position_dodge(0.2),
width = 0.2)
3 Customisations
What customisations do you see in the plots in Abbildung 10?
Code
fig_dens_colour <-
df_eng %>%
ggplot(aes(x = age_subject, y = rt_lexdec, )) +
geom_point(
color = "grey",
position = position_jitter(0.2),
alpha = 0.2) +
geom_half_violin(
aes(fill = age_subject)) +
geom_boxplot(
outlier.shape = NA,
aes(color = age_subject),
width = .3,
position = position_nudge(x=0.2)) +
labs(title = "Distribution of reaction times",
x = "Age group",
y = "LDT reaction time (ms)",
fill = "Age group") +
scale_color_colorblind() +
scale_fill_colorblind() +
theme_minimal() +
theme(legend.position = "none")
fig_point_colour <-
df_eng %>%
ggplot(aes(x = age_subject, y = rt_lexdec, )) +
geom_point(
aes(color = age_subject),
position = position_jitter(0.2),
alpha = 0.2) +
geom_half_violin() +
geom_boxplot(
outlier.shape = NA,
# aes(color = age_subject),
width = .3,
position = position_nudge(x=0.2)) +
labs(title = "Distribution of reaction times",
x = "Age group",
y = "LDT reaction time (ms)",
fill = "Age group") +
scale_color_colorblind() +
scale_fill_colorblind() +
theme_minimal() +
theme(legend.position = "none")
fig_default <-
sum_eng %>%
ggplot(aes(x = age_subject, y = mean,
colour = task, shape = task)) +
geom_point() +
geom_errorbar(aes(ymin=mean-sd,ymax=mean+sd))
fig_custom <-
sum_eng %>%
mutate(task = fct_recode(task,
"LDT" = "rt_lexdec",
"Naming" = "rt_naming"),
age_subject = fct_recode(age_subject,
"Young" = "young",
"Old" = "old")) |>
ggplot(aes(x = age_subject, y = mean,
colour = task, shape = task)) +
geom_point(position = position_dodge(0.3),
size = 3) +
geom_errorbar(aes(ymin=mean-sd,ymax=mean+sd),
position = position_dodge(0.3),
width = .3) +
geom_line(aes(group = task,
linetype = task),
position = position_dodge(0.3)) +
theme_minimal() +
labs(
title = "Reaction times per group and task",
x = "Age group",
y = "Reaction time (ms)",
colour = "Task",
shape = "Task",
linetype = "Task"
) +
theme(axis.title = element_text(size = 12,
face = "bold"),
plot.title = element_text(size = 14),
legend.title = element_text(face = "bold"))
3.1 Default themes
Firstly, theme_minimal() was added to each plot to customise the general look. There are a variety of custom themes to try, like theme_bw() or theme_classic(). Try them out.
3.2 theme()
We can also control individual components of theme by adding customisations with theme(). For example we see in Abbildung 10 A the axis titles are bolded. This was achieved by adding theme(axis.title = element_text(face = "bold)), where axis.title = indicates we want to make a change to the axis titles, element_text() indicates it’s their text that we want to change, and face = "bold" indicates we want to make the text bold. The same was done for legend.title = to make the legend title bold.
fig_no_colour + theme_minimal() +
theme(
axis.title = element_text(face = "italic")
)
Heutige Ziele 🏁
Heute haben wir gelernt, wie man…
- build multi-part plots ✅
- adjust the position of geoms ✅
- customise our plots for better data communicate ✅
4 Aufgaben
Multi-part plot. Produce Abbildung 9 and Abbildung 5 for
rt_naming(instead ofrt_lexdec). Print the plots side-by-side usingpatchwork.Labels. Use
labs()to add labels for the title, x- and y-axes, and for any aesthetics you used (shape, colour, etc.) that result in a legend. This should end with your legend title also having a custom name.Customisations. Add customisations to the two plots by choosing a default theme, followed by
theme()with adjustments for the axis titles, legend title, and plot title. You can changeface,size,family(i.e., font). You can type?themein the Console or try Googling to get some ideas. If you aren’t feeling creative, just try to replicate one of the customisations you see in Abbildung 10
Session Info
Hergestellt mit R version 4.4.0 (2024-04-24) (Puppy Cup) und RStudioversion 2023.3.0.386 (Cherry Blossom).
print(sessionInfo(),locale = F)R version 4.4.0 (2024-04-24)
Platform: aarch64-apple-darwin20
Running under: macOS Ventura 13.2.1
Matrix products: default
BLAS: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRblas.0.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
attached base packages:
[1] stats graphics grDevices datasets utils methods base
other attached packages:
[1] magick_2.8.3 gghalves_0.1.4 patchwork_1.2.0 ggthemes_5.1.0
[5] janitor_2.2.0 here_1.0.1 lubridate_1.9.3 forcats_1.0.0
[9] stringr_1.5.1 dplyr_1.1.4 purrr_1.0.2 readr_2.1.5
[13] tidyr_1.3.1 tibble_3.2.1 ggplot2_3.5.1 tidyverse_2.0.0
loaded via a namespace (and not attached):
[1] utf8_1.2.4 generics_0.1.3 renv_1.0.7 stringi_1.8.3
[5] hms_1.1.3 digest_0.6.35 magrittr_2.0.3 evaluate_0.23
[9] grid_4.4.0 timechange_0.3.0 fastmap_1.1.1 rprojroot_2.0.4
[13] jsonlite_1.8.8 fansi_1.0.6 scales_1.3.0 cli_3.6.2
[17] crayon_1.5.2 rlang_1.1.3 bit64_4.0.5 munsell_0.5.1
[21] withr_3.0.0 yaml_2.3.8 parallel_4.4.0 tools_4.4.0
[25] tzdb_0.4.0 colorspace_2.1-0 pacman_0.5.1 vctrs_0.6.5
[29] R6_2.5.1 lifecycle_1.0.4 snakecase_0.11.1 bit_4.0.5
[33] htmlwidgets_1.6.4 vroom_1.6.5 pkgconfig_2.0.3 pillar_1.9.0
[37] gtable_0.3.5 glue_1.7.0 Rcpp_1.0.12 xfun_0.43
[41] tidyselect_1.2.1 rstudioapi_0.16.0 knitr_1.46 farver_2.1.1
[45] htmltools_0.5.8.1 labeling_0.4.3 rmarkdown_2.26 compiler_4.4.0 Literaturverzeichnis
Nordmann, E., McAleer, P., Toivo, W., Paterson, H., & DeBruine, L. M. (2022). Data Visualization Using R for Researchers Who Do Not Use R. Advances in Methods and Practices in Psychological Science, 5(2), 251524592210746. https://doi.org/10.1177/25152459221074654
Wickham, H., Çetinkaya-Rundel, M., & Grolemund, G. (2023). R for Data Science (2. Aufl.).